Translatomics study of chinese hamster ovary cell cultures

106 Chapter 5: Understanding Translational Control Mechanisms of the mTOR Pathway in CHO Cells by Polysome Profiling .... Control of translation activity by the mTOR pathway in CHO cell

TRANSLATOMICS STUDY OF CHINESE HAMSTER OVARY CELL CULTURES

DEPARTMENT OF CHEMICAL AND BIOMOLECULAR ENGINEERING

NATIONAL UNIVERSITY OF SINGAPORE

2013

(Blank Page)

DECLARATION

I hereby declare that the thesis is my original work and it has been written by

me in its entirety I have duly acknowledged all the sources of information

which have been used in the thesis

This thesis has also not been submitted for any degree in any university

previously

Franck Courtes

20 June 2013

(Blank Page)

Acknowledgements

This thesis would not have been possible without the guidance and the help of several individuals who in one way or another contributed and extended their valuable assistance in the preparation and completion of this PhD

First and foremost, I would like to express my deepest appreciation to Professor Miranda Yap, Assistant Professor Lee Dong Yup and Doctor Niki Wong for their insightful comments, persistent help and inspiration as I hurdle all the obstacles in the completion of this thesis I sincerely wish a quick recovery to Professor Yap

My sincere thanks also go to Assistant Professor Tong Yen Wah, Associate Professor Ren Ee Chee and Associate Professor Lee Yuan Kun members of

my thesis Advisory Committee Their continuous support, instructive advices and motivating discussions rendered for the past four years have been instrumental to my research

I would like to acknowledge my collaboration with Doctor Bernard Loo and his research officers Sze Wai Ng and Hsueh Lee Lim from the microarrays group of BTI, for guiding the design of microarrays experiments, which were crucial for my PhD

Thanks to Doctor Andrea Camattari, I had an opportunity to collaborate with Doctor Leah Vardy, whose contribution to my PhD had been invaluable I am indebted to Leah for welcoming me to her lab, for training me to perform polysome profiling and for always fostering my motivation in research

Additionally, I would like to acknowledge Kristin Chong Peini, a research officer from Leah’s group who perfected my polysome profiling skills

My sincere gratitude goes to Doctor Muriel Bardor for her indispensable advices and encouragement Muriel has always been not only my great mentor but also a wonderful friend I would like to thank her for being available to

discuss about sciences as well as personal life and especially for welcoming

me to her family in time of need

Many thanks to Doctor Yuan Sheng Yang and Doctor Say Kong Ng for their kind consideration and their time to discuss about my project and share with

me their valuable recommendations

My gratitude goes to Professor Dedon and his graduate student Chen Gu who were kind enough to accept to measure the pseudouridylation content of my RNA samples in their laboratory in the Massachusetts Institute of Technology

I also am very thankful for Professor Wei-Shou Hu for his lecture and discussions on CHO cell culture that enlightened my PhD project and research motivations

Many thanks to the human resources staffs of BTI, particularly Mei Yi Wong and Siew-Chin Chung, for their persistent and prompt assistance on all the administrative matters during the past years

My appreciation also goes to Doctor Vincent Vagenende, Danny Ong, U-ming Lim and Doctor William Chong for having been such great and supportive friends I will always cherish the memories of our joyful times that we spent together in Singapore

My everyday life in the animal cell technology group would not have been as pleasant without my PhD colleagues including Janice Tan, Steven Ho and Wanping Loh Thank you for always being willing to discuss technical details and for being my friends Very special thanks go to Janice for all of the special treats and fruits that she shared with me, when I became evidently stressful with my thesis

I would like to acknowledge Yasotha Kathiraser, whose endless kindness, simplicity and happiness in life has been a wonderful life lesson to me

My work on this thesis was supported by the Agency for Science Technology and Research that provided me with the Singapore International Graduate Award, the Bioprocessing Technology Institute that provided me with necessary facilities for carrying out this research work, and the National University of Singapore.

Last but not least, I am heartily thankful to my beloved wife for her endless love, care, understanding, moral support and inspiration during my research Thank you for showing me what love means every day I am also thankful to

my family especially my parents Alain and Genevieve as well as my beloved aunts and uncles, who sacrificed a lot to provide me with a good education and

a loving environment to grow up in I am grateful for their infinite love and support throughout everything in my life

It has been an immense honor to spend this part of my life in the great city of Singapore and I am looking forward to keeping strong ties with this country

Table of Contents

Acknowledgements iii

Table of Contents vi

Summary xi

List of Tables xiii

List of Figures xiv

Chapter 1: Introduction 1

1.1 Background 1

1.2 Thesis objective 3

1.3 Thesis organization 4

Chapter 2: Literature Review 6

2.1 Translatome: the missing gap between transcriptome and proteome 6

2.1.1 Transcriptome: “What seems to happen” 10

2.1.2 Proteome: “What makes it happen” 14

2.1.3 Discrepancy between transcriptome and proteome 17

2.1.4 Translatome: “What cells need to happen” 19

2.2 Translational control 22

2.2.1 Reasons for regulating translational activity 22

2.2.2 Molecular mechanisms of translation 23

2.2.3 Initiation of translation 25

2.2.4 Different mechanisms of translational control 28

2.3 mTOR signaling pathway 32

2.3.1 Organization of mTOR pathway 32

2.3.2 Upstream signaling of mTOR activity 34

2.3.3 Downstream targets of mTOR pathway 35

2.3.4 Rapamycin treatment of mTOR pathway 39

2.4 Small Nucleolar RNAs (snoRNAs) 41

2.4.1 Biogenesis and mode of action of snoRNAs 42

2.4.2 Function of H/ACA box snoRNAs 45

Chapter 3: Materials and Methods 47

3.1 Cell culture 47

3.1.1 Cell lines 47

3.1.2 Cell maintenance 48

3.1.3 Batch cultures 49

3.1.4 Feeding of batch cultures 49

3.1.5 Rapamycin treatment of batch cultures 50

3.1.6 Determination of cell viability 50

3.1.7 Determination of monoclonal antibody titers 50

3.1.8 Measurement of residual glutamine and glucose concentrations 51

3.2 Translation activity 51

3.2.1 Sucrose solutions and gradient preparation 51

3.2.2 Polysome extraction 52

3.2.3 Polysome profiling and fractionation 53

3.2.4 Fraction pooling 54

3.2.5 RNA extraction from pools and purification 55

3.3 Translatome profiling and transcriptome profiling 56

3.3.1 Preparation of labeled cDNA library for microarrays 57

3.3.2 Microarrays procedure 58

3.3.3 Microarray data analysis 60

3.4 Targeted relative quantification of specific RNA level 61

3.4.1 Primers 61

3.4.2 Template DNA 64

3.4.3 Quantitative-RT PCR 70

3.5 mTOR pathway activity 71

3.5.1 Total protein extraction 71

3.5.2 Total protein quantification 71

3.5.3 Immunobloting 71

3.6 Generation of stable CHO U19 pools 72

3.6.1 Construction of expression vector pcDNA3.1/hygro-U19 72

3.6.2 Transfection of CHO M250-9 cell line 78

3.6.3 Selection of stable CHO U19 pool 80

3.7 Quantification of pseudouridine in 28S rRNA 80

3.8 Design of experiment and calculation of variables effect 82

Chapter 4: Developing a Strategy for the Translatomic Analysis of CHO Cells 83

4.1 Introduction 83

4.2 Results and discussion 84

4.2.1 Generation of first translatome data in CHO cells 84

4.2.2 Global translation activity during exponential growth phase 93

4.2.3 Translatome for identifying key growth genes 95

4.2.4 Translational control mechanisms in CHO cells 101

4.3 Summary 106

Chapter 5: Understanding Translational Control Mechanisms of the mTOR Pathway in CHO Cells by Polysome Profiling 107

5.1 Introduction 107

5.2 Results and discussion 109

5.2.1 Targeted inhibition of the mTOR pathway by rapamycin 109

5.2.2 Maintenance of mTOR pathway activation through nutrient supply

117

5.2.3 Control of translation activity by the mTOR pathway in CHO cell cultures 121

5.3 Summary 127

Chapter 6: Characterizing the Interplay between mTOR Pathway and snoRNA U19 129

6.1 Introduction 129

6.2 Results and discussion 130

6.2.1 Regulation of snoRNA U19 expression by mTOR pathway 130

6.2.2 Model for inducible pseudouridylation in CHO cell cultures 138

6.3 Summary 149

Chapter 7: Conclusions and Recommendations 150

7.1 Conclusions 150

7.2 Recommendations for future work 153

7.2.1 Inclusion of proteome data 153

7.2.2 Updated functional annotation of CHO genes 153

7.2.3 Cell engineering on suggested key growth targets 154

7.2.4 Nutrient availability control 154

7.2.5 Single cell cloning of CHO-U19 155

7.2.6 Measurement of targeted pseudouridylation sites 155

7.2.7 Microarrays profiling of snoRNAs in CHO cells 156

Abbreviations 157

Bibliography 160

Appendices 176

Appendix A - List of publications 176

Appendix B - RNA quality assessment via bioanalyzer prior to cDNA library synthesis 177 Appendix C - cDNA library quality assessment via bioanalyzer 180 Appendix D - Standard curves for primers efficiency 183 Appendix E - Receipt of RNA samples shipment for pseudouridylation analysis in Professor Dedon’s laboratory 188

Summary

Over the past decade, the transcriptomics and proteomics profilings of Chinese hamster ovary (CHO) cells have been extensively used for understanding cellular mechanisms and for identifying cell engineering targets towards an optimization of cell cultures However, transcriptome and proteome reports have described limited success partially due to un-elucidated translational control mechanisms

In order to address this knowledge gap, we developed and applied a resolution translatomic platform based on the combination of polysome profiling and microarray technologies, which measured the translational efficiency of every gene on a global scale With this platform we were able to successfully generate the first translatome data of exponentially growing CHO cells and suggested highly and stably translated genes as potential key growth genes for genetic engineering Furthermore, correlation analysis between translatome and transcriptome data indicated that more than 90% of the genes were potentially subjected to translational control mechanisms, notably regulated by the mammalian target of rapamycin (mTOR) pathway

high-The evaluation of the impact of the mTOR pathway on translation activity in CHO cells revealed that the starvation state in batch cultures had similar effect

as rapamycin inhibition causing a shift of polysomes towards monosomes and

a decrease in cellular growth Conversely, nutrient supplementation in batch cultures was able to maintain the activity of the mTOR pathway as well

fed-as a 2-fold higher global translation activity thereby extending cellular growth

by 5 days and increasing final recombinant protein titer by 4-fold The mTOR pathway appeared thus to influence the fate of CHO cell cultures via translational control mechanisms which were further investigated by comparing the translatome of rapamycin-treated versus untreated control with our translatomics platform

Upon rapamycin treatment, the level of short nucleolar RNA U19 (snoRNA U19), which guides the two most conserved pseudouridylation modifications

on 28S ribosomal RNA (rRNA) was found to be reduced by 2-fold A design

of experiment (DOE) analysis of snoRNA overexpression in CHO cells provided the first insight into a potential mTOR pathway regulation mechanisn

of translation activity via inducible rRNA modification and a model was proposed

These findings demonstrate how a translatomics-based strategy enabled the identification of key cell growth genes in recombinant CHO cultures, thereby guiding to the discovery of a new mechanism of the crucial mTOR pathway which paves the way for future improvement of CHO cell cultures

List of Tables

Table 2.1: Survey of all reported CHO transcriptome and proteome analysis 9

Table 3.1: Parameters for microarrays scanning with GenePix 4000B scanner 60 Table 3.2: Parameters for qRT-PCR primer design 62 Table 3.3: qRT-PCR primer sequences 62

Table 3.4: Composition of reaction mixture for PCR with Taq polymerase (Fermentas) 63 Table 3.5: Standard master mix composition for gDNA digestion 66 Table 3.6: Standard master mix composition for polyadenylation reaction 68 Table 3.7: Oligo dT priming mixture for reverse transcription 69 Table 3.8: Reaction mixture for reverse transcription 69 Table 3.9: Master mix for qRT-PCR reaction 70

Table 3.10: Composition of reaction mixture for PCR with high fidelity Platinum Taq polymerase (Invitrogen) 74

Table 3.11: Composition of reaction mixture for digestion with NheI and BamHI restriction enzymes 75 Table 3.12: Composition of reaction mixture for phosphatase treatment 76

Table 3.13: Composition of reaction mixture for ligation of insert U19 in backbone pcDNA3.1/Hygro 76 Table 3.14: Parameter settings used for FACS 79 Table 4.1: Functional enrichment of genes with constant and high translational efficiency during the exponential growth phase 100

Figure 2.5: Major molecular events that lead to cap-dependent translation initiation 26 Figure 2.6: IF2 translational control mechanism 28 Figure 2.7: 4EBP translational control mechanism 31 Figure 2.8: Simplified overview of the mTOR pathway in mammalian cells 32 Figure 2.9: Structural organization of the mTOR protein 33

Figure 2.10: Activation of IF4B by the downstream effector of mTOR pathway S6K1 37

Figure 2.11: Chemical structure difference between uridine and pseudouridine 42 Figure 2.12: Biogenesis of snoRNAs 43 Figure 2.13: H/ACA snoRNPs and mode of action 44

Figure 3.1: Map of the expression vectors co-transfected in CHO M250-9 cell line 47 Figure 3.2: Polysome profiling and global translation 53 Figure 3.3: Polysome profile fractionation and pooling 54 Figure 3.4: Organigram overview of the transcriptome profiling and translatome profiling 57

Figure 3.5: Mixer and glass slide assembly of the 12-plex microarrays 59

Figure 3.6: Workflow overview for the synthesis of cDNA template 65

Figure 3.7: Schematic overview of the cloning procedure for the construction of pcDNA3.1-hygro-U19 expression vector 72

Figure 3.8: Primer design for PCR amplification of the insert-U19 73

Figure 3.9: Determining the effect of the variable (X) on the output (Y) for an example of design with two variables 82

Figure 4.1: Schematic overview of the translatomic platform 85

Figure 4.2: Polysome fractionation and resolution of translatome data 86

Figure 4.3: Box plot of pre-normalized microarray data 90

Figure 4.4: Correlation inter pre-normalized microarray data sets 92

Figure 4.5: Validation of the microarray data by qRT-PCR 93

Figure 4.6: Framework of translatome analysis for the identification of key growth genes 94

Figure 4.7: Workflow of translatome data processing towards the identification of key growth genes in CHO cells 97

Figure 4.8: Correlation between transcript level and translational efficiency 102

Figure 5.1: Kill curve for rapamycin in CHO-mAb cultures 110

Figure 5.2: Effect of rapamycin of mTOR pathway activation level 112

Figure 5.3: Effect of rapamycin treatment on CHO-mAb cell culture 113

Figure 5.4: Effect of rapamycin treatment on global translation 115

Figure 5.5: Feeding nutrients enhances growth performances of batch cultures 119

Figure 5.6: Extension of mTOR pathway activation by nutrient supplementation 120

Figure 5.7: Effect of mTOR pathway on global translation activity upon

nutrient supplementation 122

Figure 5.8: Viable cell density: a balance between cellular growth and death 124

Figure 5.9: HC and LC mRNAs translational efficiency upon nutrient supplementation and mTOR pathway activation 126

Figure 6.1: Identification of genes significantly affected by polysome shift upon rapamycin treatment 131

Figure 6.2: SNHG4 is a non protein coding host gene 134

Figure 6.3: Characterization of the intron region in SNHG4 135

Figure 6.4: Identification of snoRNA U19 hosted in SNHG4 of CHO cells 136 Figure 6.5: snoRNA U19 level was reduced upon rapamycin treatment in CHO cells 137

Figure 6.6: Design of the snoRNA U19 expression vector 139

Figure 6.7: Transfection efficiency of stably transfected CHO cells 140

Figure 6.8: Hygromycin kill curve 141

Figure 6.9: Selection of stable CHO U19 pool at 300 µg.mL-1 hygromycin 142 Figure 6.10: Overexpression of snoRNA U19 in CHO-U19 pool 143

Figure 6.11: Design of experiment to determine the effects of snoRNA U19 and rapamycin treatment on cellular growth and pseudouridylation 145

Figure 6.12: Model for mTOR pathway control of growth via inducible pseudouridylation 148

Chapter 1: Introduction

1.1 Background

Chinese hamster ovary (CHO) cells are one of the most commonly used mammalian host cell lines by biopharmaceutical industries for the production

of recombinant proteins of therapeutic interest (Jayapal et al 2007) With over

116 recombinant proteins approved by the Food and Drug Administration since the 80s, for the treatment of diseases like cancer and arthritis (Aggarwal 2007), biopharmaceuticals represent today a key market that generated 53.8 billion US$ sales revenue in 2011 in the USA (Aggarwal 2012) This economical success has also strongly benefited the local Singapore economy

by providing several hundreds of jobs for the production of recombinant proteins (Wong and Yap 2007)

In order to meet the increasing demand for such recombinant proteins, effort have been made for significantly enhancing the production capacity of CHO cell cultures (Barnes and Dickson 2006) Some of the major improvements were achieved by empirically designed strategies (Kim et al 2012a; Lim et al 2010b; Wurm 2004) without clear understanding of the underlying cellular mechanisms The lack of such crucial understanding has hindered the development of knowledge-based strategies to fully optimize and control cell culture processes To address this limitation, several ‘‘-omics’’ profiling technologies such as transcriptomics and proteomics have been successfully utilized to gain a more in-depth insight into these cellular mechanisms (Kuystermans et al 2007; Omasa et al 2010)

Interestingly, studies of the transcriptome and/or proteome of CHO cells under various conditions have commonly reported a general lack of correlation between mRNA and proteins levels (Nissom 2006) This was also observed in various human cell lines and yeast (de Nobel et al 2001; Gygi et al 1999; Pradet-Balade et al 2001) Therefore, some translational control mechanisms seemed to be determinant in regulating the flow of information between mRNA and proteins levels in cells (Pradet-Balade et al 2001)

Cells have evolved translational control mechanisms in order to adjust their translation activity, which is an energy-intensive process (Shimizu et al 2001), with respect to the resource availability Hence by controlling translation, cells ensure an appropriate coupling of cell growth and metabolism with their surrounding environmental conditions (Dethlefsen and Schmidt 2007) One of the major regulators of translational control is the mammalian target of rapamycin (mTOR) signaling pathway (Ma and Blenis 2009)

Based on the documented and promising potential of the mTOR pathway towards enhancing growth (Sarbassov et al 2005; Schmelzle and Hall 2000), Dressen and Fussenegger (2011) recently exploited the mTOR pathway as a cell-engineering target Over-expression of a human TOR complex in CHO cells transiently expressing monoclonal antibodies led to increased growth, viability and productivity in micro-carrier type of culture However, in that report, the mTOR pathway activity was not assessed and it remained unknown how increasing the level of TOR complex actually contributed to the

enhancement of cell phenotype Furthermore, altering the level of mTOR pathway players may only partially affect phosphorylation cascade events, the actual molecular signals of the mTOR pathway As a result, the molecular mechanisms of the mTOR pathway in CHO cells need to be investigated in order to fully decipher its role in bioprocessing relevant characteristics As a matter of fact, Lee and Lee (2012) recently demonstrated that the mTOR pathway extended cellular viability via autophagy induction after rapamycin treatment thereby illustrating the importance of understanding the mTOR pathway’s molecular mechanisms in CHO cell cultures

These observations highlight the necessity to comprehend the translational control of genes via the mTOR pathway in CHO cells, as that may direct us towards the design of strategies to enhance bioprocesses for improved recombinant protein production

1.2 Thesis objective

The objective of the thesis is to understand translational control mechanisms

in CHO cells and consequently investigate new strategies for improving cell growth via a “translatomic” approach

The scope involved:

1 Establishing the first translatome map of genes which support cellular growth of CHO cells as a novel approach to (A) discover and propose cell engineering targets as well as (B) to establish the state of correlation between transcriptome and translatome (chapter 4)

2 Evaluating the impact of mTOR pathway translational control on growth and productivity of CHO cell cultures upon rapamycin inhibition and nutrient supply induction (chapter 5)

3 Elucidating new players of mTOR pathway mediated translational control

by analyzing translatome profile of rapamycin treated CHO cells and by performing functional cell engineering on identified snoRNA U19 target (chapter 6)

1.3 Thesis organization

The thesis consists of a total of 7 chapters

Chapter 1 introduces the context of the thesis and details the objective and

scopes

Chapter 2 presents a review of literature on four areas: (1) transcriptomic and

proteomic landscape in CHO cells evidencing the existence of discrepancies due to translational control mechanisms that can be adequately addressed with translatomics as reported for other organisms; (2) general translational control

in eukaryotes with a focus on mTOR pathway in (3); as well as (4) snoRNAs biology and relevance to the mTOR pathway translational control

Chapter 3 details the materials and methods used

Chapter 4 describes the development of a polysome profile platform resulting

in the first CHO translatome data and proposes cell-engineering targets

Chapter 5 reports the evaluation of the mTOR pathway impact on translation

regulation in CHO cell cultures and how it affects growth and productivity

Chapter 6 documents the identification and subsequent engineering of

snoRNA U19 as a potential effector of the mTOR pathway translational control mechanisms, which led to the suggestion of a biological model

Chapter 7 summarizes the important conclusions resulting from this thesis

and provides recommendations for future work

Chapter 2: Literature Review

Firstly, this literature review provides the state of the art of transcriptomics and proteomics in CHO cells as an indication of strong scientific interest in systems-level understanding as well as an evidence of discrepancies between these two “-omics” due to translational control mechanisms Translatomics solutions, like the one applied in chapter 4, to study translational control are also reviewed Secondly, the general context of translational control, which regulates the translatome in eukaryotes, is described and leads to the third point focusing on the mTOR pathway mechanisms that were specifically investigated in chapter 5 Finally, the fourth point provides biological backgrounds on snoRNAs and reviews the increasing amount of evidence of snoRNAs being involved in gene expression regulation and possibly related to the mTOR pathway, where a mechanistic model was suggested in chapter 6

2.1 Translatome: the missing gap between

transcriptome and proteome

Proteins are produced from the genetic information (DNA) via the two

biosynthetic steps known as transcription and translation (Figure 2.1)

Transcription takes place in the cellular nucleus by RNA polymerases After post-transcriptional processing events such as splicing that removes non-coding regions (Black 2003) and addition of a 7-methylguanosine cap in front

of the 5’ end of the mRNA (5’cap; Kapp and Lorsch 2004), the mRNA is transported from the nucleus to the cytoplasm where translation takes place Once formed, proteins may undergo post-translational modifications such as

phosphorylation or glycosylation before being used inside cells, secreted or degraded (Walsh et al 2005), which results in the overall cellular phenotype Since the past decade, functional genomics has been increasingly utilized to describe this complex relationship between genome and phenotype in cellular systems (Borrebaeck 1998; Evans et al 1997) Various “-omics” technologies have been employed to support functional genomics by characterizing the transcriptome and proteome in cells (Geschwind and Konopka 2009; Glinski

and Weckwerth 2006), which is also known as the ‘-omics’ cascade (Figure

2.1) This enables the understanding of the multi-leveled cellular regulation in

response to stimuli and environmental or genetic

Figure 2.1: Biosynthesis of protein in eukaryotic cells and corresponding omics” cascade

“-Translatome:

“What cells need to happen”

mRNA Transcription

The typical experimental design for “-omics” studies in CHO cells compare cellular states between a control and an altered condition inducing a desired phenotypic change at the transcript and protein expression levels, respectively (Kuystermans et al 2007; O'Callaghan and James 2008) The altered condition can be a different growth phase, the feeding of supplements, a decrease in culture temperature or genetic modifications Thereafter, the differentially expressed transcripts or proteins due to the altered condition are assumed to be involved in cellular mechanisms leading to the observed phenotype change

Figure 2.2 summarizes the growing number of publications related to the

proteome and transcriptome of CHO cells

Figure 2.2: Trends of the cumulative number of publication reporting transcriptome and/or proteome data in CHO cells

(Only transcriptome based on microarray technology was considered)

The first systems-level analysis of a CHO cell line was a proteomic study performed by Lee et al (1996) using two dimensional electrophoresis, in order

to decipher the protein expression after stimulation by hormones and growth

CHO proteomic analysis CHO transcriptomic analysis

factors such as insulin and fetal calf serum Interestingly, the first transcriptome data for CHO cells was reported only ten years later by Baik et

al (2006) describing the molecular response of cells to temperature shift from

37 to 30°C To date, there are a total of 11 and 22 transcriptomic and proteomic CHO analyses respectively This steady increase underscores the strong interest of the bioprocessing community towards systems-level understanding The next two sections will describe the literature of

transcriptome and proteome analysis in CHO cells (Table 2.1)

Table 2.1: Survey of all reported CHO transcriptome and proteome analysis

Lee et al (1996)

Growth factor supplements

20 ng.mL-1 fibroblast growth factor

or 1 pg.mL-1 insulin or 10% calf serum - + Champion et al (1999) Exponential growth phase - + Kaufmann et al (1999) Temperature shift (37 to 30°C) - + Lee et al (2003) Hyperosmotic pressure (450 mOsm.kg-1) - + Van Dyk et al (2003) Butyrate (0.5 mM) and zinc sulphate (80µM) treatment - + Hayduk et al (2004) 70–80% confluence cells - + Hayduk et al (2005) Clones selected at 0, 20, 200, 640, and 5120 nM methotrexate - + Baik et al (2006) Temperature shift (37 to 33°C) + +

Li et al (2006) Dimethyl Sulfoxide (DMSO) treatment (1.5% v/v) - + Nissom et al (2006) High vs low GFP producers + + Wong et al (2006) Batch vs fed-batch cultures + -

De Leon Gatti et al

(2007) Butyrate treatment (0.5, 1, and 2 mM) + - Pascoe et al (2007) Clone consuming vs Clone not consuming lactate - + Doolan et al (2008) Control vs PACE transfected cells

Table 2.1 continued

Carlage et al (2009) High vs low producers - + Yee et al (2009) Temperature shift (37 to 33°C) + - Kantardjeff et al (2010)c Butyrate treatment (2 mM) at 33°C + -

Doolan et al (2010) Fast vs slow growth rate cells + + Klausing et al (2011) Butyrate treatment (2 mM) + - Kim et al (2011) Hydrolasate supplements to serum free

Meleady et al (2011) Stable productivity clones vs unstable productivity clones - +

Doolan et al (2012) Stable productivity clones vs unstable

2.1.1 Transcriptome: “What seems to happen”

Transcriptome is a quantitative measurement of transcripts on a global scale at

a given time point via microarrays Microarrays are arrays that contain a large number of probes (typically 10 to 30 thousands) which target specific transcripts by means of pairing hybridization (Gresham et al 2008) The number and specificity of probes on arrays has dramatically increased over the past decade thanks to the progress of probe synthesis and microarrays technologies as well as the sequencing advancements of the CHO genome (Jacob et al 2009) For example, the first reported CHO transcriptome data (Baik et al 2006) were actually generated with a rat specific microarrays of 5,029 probes and a mouse specific microarrays of 7,140 probes targeting 1,655

and 4,643 known genes respectively Because of the partial sequence homology between rat/mouse and CHO genes (Yee et al 2008b), the data analysis was limited to general cellular process groups Shortly after, Nissom

et al (2006) synthesized the first reported CHO specific microarrays containing 14,592 cDNA probes based on Expressed Sequence Tag (EST; Wlaschin et al 2005) targeting 7,559 unique CHO genes, for the comparison between high and low Green Fluorescent Protein (GFP) producers Seventy-seven transcripts were observed as differentially regulated of which 41 had known annotated functions including up-regulation of transcripts coding for the enzymes responsible for opening the DNA HMGN3 and HMGB1, as well

as for the ribosomal protein RPS2 and RPS27 of the 40S subunit of ribosomes

Using similar CHO-EST microarrays, a targeted transcriptomic approach was used to track the transcript regulation accompanying the onset of apoptosis between batch and fed-batch cultures (Wong et al 2006a) Seven core

apoptotic genes were detected as differentially expressed including, Faim,

Agl2, Fadd and Requiem These genes were then engineered to successfully

enhance the apoptosis resistance of CHO cells (Lim et al 2010a; Wong et al 2006b)

Butyrate treatment, an histone deacetylation inhibitor often utilized as additive

to stimulate recombinant protein production by improving gene accessibility (Jiang and Sharfstein 2008), altered the expression of 25 genes notably involved in apoptosis and protein folding (De Leon Gatti et al 2007) Another study which identified 742 differentially expressed transcripts upon butyrate

treatment confirmed the importance of apoptosis and protein folding functional clusters as well as cell cycle, protein transport and lipid metabolism (Yee et al 2008a) The transcriptome of CHO cells with increased specific productivity resulting from butyrate treatment and a temperature shift at 33°C showed 900 differentially regulated genes largely enriched in the secretory pathway including the Golgi apparatus, cytoskeleton protein binding and small GTPase-mediated signal transduction functional clusters (Kantardjieff et al 2010) During a study involving temperature shift only, 237 transcripts were altered with significant up-regulation in protein trafficking (including genes

Kpna3, Rab5A, Gga3, Clta) and cytoskeleton re-organization (Yee et al

2009) This lower number of altered transcripts due to temperature shift alone

as compared to the combined effects of butyrate and temperature seemed to indicate that butyrate has a greater impact on genes expression Of interest,

there were seven genes including Pitpna, Timm8b, Napg, Gga3, Arl1, Vdp and

Ap1g1, displaying analogous changes between Yee et al (2008a) and

Kantardjeff et al (2010) studies Similarly, thirty-five genes were common in both Yee et al (2008a) and Yee et al (2009) transcriptome data These genes are likely related to productivity of CHO cells since they were common to both productivity-increasing strategies of temperature shift and/or butyrate treatment The latest publication on butyrate treatment reported a total number

of 1,461 transcripts differentially regulated with the 10 highest fold change enriched in transcription and translation, cell cycle and apoptosis (Klausing et

al 2011)

The effect of cell engineering on transcriptome regulation was also studied by comparison between an engineered cell line overexpressing the Paired Amino acid Cleaving Enzyme (PACE) and the untransfected control PACE is known

to increase the rate of post translational processing and hence increase productivity (Roe et al 2004) Transcriptome data indicated that 1,076 transcripts were differentially expressed with most significant impact on the endoplasmic reticulum and the Golgi apparatus (Doolan et al 2008), which was in agreement with the expected role of PACE The same authors reported two other studies (Doolan et al 2012; Doolan et al 2010) where they compared fast and slow growing cell lines to identify genes which drive fast growth; and stable and unstable clones to identify genes for high stability High growth rate phenotype in CHO cells was observed to correlate with the differential expression of 118 annotated transcripts (Doolan et al 2010) while long term maintenance of productivity was reported to involve the differential expression of 19 genes (Doolan et al 2012)

In the recent years, along with the development of high throughput sequencing technologies, a new approach to directly sequencing transcripts instead of probing them on microarrays has been employed to generate transcriptome data (Becker et al 2011; Jacob et al 2010) However, this approach, still in its infancy poses bioinformatic challenges such as the need to properly assemble such high number of sequence reads As a result, the more robust and established microarrays technology was used for this thesis

2.1.2 Proteome: “What makes it happen”

Proteome is a quantitative measurement of protein level on a global scale through proteomics Proteomics is the second main “-omics” together with transcriptomics, which has attracted growing interest in CHO cells over the past decade This section reviews the key proteomics contributions which are important to understand the general context of the “-omics” family in CHO cells to which translatomics was integrated

After the first exploration of the proteome which was based on identification of proteins (Lee et al 1996), Champion et al (1999) established

immuno-a primimmuno-ary CHO specific mimmuno-ap of 25 proteins Such mimmuno-apping efforts immuno-are necessary for accurate identification of proteins via the high throughput mass spectrometry (MS) technique, and was further extended with 224 (Hayduk et

al 2004) and 179 (Lee et al 2010) proteins identified by MS/MS Currently, with the recent sequencing of the full CHO-K1 genome (Xu et al 2011), the capacity of protein identification was increased by 8-fold (Baycin-Hizal et al 2012)

A shift of temperature to 30°C was observed to alter the quantity and postranslational modification of 11 unidentified proteins (Kaufmann et al 1999) A similar temperature shift to 33°C significantly modified the quantity

of 9 identified proteins involved in protein folding and metabolism out of a total of 26 identified proteins (Baik et al 2006) Six proteins including VIM, GAPDH, LGALS1, ACTB, PHB and TPII, were also identified in another proteomic analysis involving temperature shift to 31°C which identified a total

of 23 differentially regulated proteins (Kumar et al 2008) Increase in osmotic pressure at 450 mOsm.kg-1, which induced a slower cellular growth but a concomitant augmentation of specific productivity (from 15 to 35 pg.cell-

1.day-1), correlated with the up-regulation of the proteins GAPDH and pyruvate kinase (Lee et al 2003) Nine years later, the same research team, looked again at the effect of osmotic pressure but with the addition of betaine

as an osmoprotectant to minimize the decrease of cellular growth, and could identify 16 proteins differentially regulated including up-regulation of GAPDH and a pyruvate kinase (Kim et al 2012b) as observed in their previous study (Lee et al 2003) These two identified proteins probably led to increased metabolic energy for recombinant protein production

Butyrate treatment induced the expression of four proteins GRP75, ENO, C1 and TXN supporting an increase in metabolic requirement of CHO cells (Van Dyk et al 2003) On the other hand, the chemical Dimethyl Sulfoxide (DMSO) which was shown to enhance CHO cells productivity (Liu et al 2001), led to the down-regulation of 6 identified proteins mainly related to glycolysis and protein folding including triosephosphate isomerase (TIP), glyceraldehyde 3-phosphate hydrogenase (GAPDH) and aldolase (ALDO) The decreased quantity of these enzymes implied that DMSO could alter the distribution of substrate metabolism towards recombinant protein production (Li et al 2006b) Moreover, the cellular mechanisms underlying gene amplification with methotrexate, a major selection drug for Dihydrofolate Reductase (DHFR) expression system in CHO cells, were monitored at the proteomic level (Hayduk and Lee 2005) Seventeen proteins involved in

translation, energy pathways, chaperons and cytoskeletal proteins were differentially regulated

Overexpression of PACE in cells changed the quantity of 60 identified proteins which were involved in chaperone activity, protein folding, assembly and secretion as well as protein translation (Meleady et al 2008) In a proteomic analysis of lactate-consuming CHO cells, glycolytic enzymes were differentially expressed and supported the metabolic shift as well as proteins related to cell structure and protein processing (Pascoe et al 2007)

In a proteomic analysis of CHO cells engineered with the apoptosis inhibitor Bcl-XL, 32 proteins were differentially expressed and were involved in protein metabolism, transcription and cytoskeleton functional clusters (Carlage

et al 2009) In another study, twelve differentially expressed proteins involved

in glucose metabolism, protein translation and folding were identified through comparison between high and low producer CHO cells (Meleady et al 2011) Proteomic analyses across the exponential and the stationary growth phases were used to monitor the dynamic of proteins associated with growth and apoptosis Fifty nine proteins were identified including molecular chaperone and isomerases GRP78 and PDI and cell growth markersMCM2 and MEM5 with dynamic changes (Carlage et al 2012)

Proteins are the drivers of biochemical reactions and therefore are major effectors of the cellular phenotype The proteome provides information on what enables the phenotype while the transcriptome shows what genes seem to

be expressed in cells This distinction highlights the utility of the proteome approach, which is able to specifically detect the final gene product rather than

an intermediate mRNA product as stated by Van Dyk et al (2003)

2.1.3 Discrepancy between transcriptome and proteome

The degree of connection between transcriptome and proteome has been a topic of interest to the research community

The first attempt at correlating transcriptome and proteome of CHO cells indicated that the 6 proteins significantly up-regulated (by a factor of 2.2 to 6.7) after temperature shift at 33°C and which had matching probes on the microarrays, had no linear correlation with their respective transcripts whose levels were little or not significantly altered (Baik et al 2006) Although the technical limitation of microarrays and proteomic platforms have contributed

to some extent to this apparent weak correlation between transcriptome and proteome, this lack of correlation continued to persist even when microarray and proteomic technologies were improved In fact, in addition of showing differing extent of expression change, some protein expression levels were opposite in direction compared to the transcript levels in a combined analysis

of CHO specific microarrays and proteome data of high producer CHO cells (Nissom et al 2006) The cross analysis of the transcriptome (Doolan et al 2008) and proteome (Meleady et al 2008) of a same PACE-engineered cell line showed that for 21 identified proteins which had corresponding probes on the microarrays, there were three cases of opposite change direction but no quantitative information was provided for the remaining 18 In another

analysis, on the effect of butyrate treatment, out of the 7 proteins differentially regulated that had corresponding probes on the microarrays, only 3 showed similar trend while the other 6 were unchanged at the transcript level (Yee et

al 2008a) Similarly, among the 21 genes that were identified differentially regulated both at the protein and transcriptome levels through comparison of fast and slow growth rate cells, only 14 genes showed common direction change between transcriptome and proteome (Doolan et al 2010) A combined analysis of two separate publications that compared the same stable and unstable productivity clones at the transcriptome (Doolan et al 2012) and proteome (Meleady et al 2011) levels, clearly seems to indicate that there is not a single identified differentially regulated gene that showed correlation

These findings imply that a change of a transcript level upon any common environmental trigger may not necessarily be accompanied by a similar change (direction and amplitude) at the protein level In fact, it has been estimated that only 20 to 40% of the protein concentrations are determined by the corresponding transcript level in mammalian cells (Cox et al 2005; Tian et

al 2004) which suggest that not all mRNA are translated with the same efficiency This led to the concept of translation on demand introduced by (Brockmann et al 2007) whereby there is permanently an existing pool of mRNA present in the cytoplasm and cells decide what transcript to translate based on their needs Thus, for a more inclusive interpretation of the “-omics” landscape in CHO cells, deeper understanding of translational regulation is required The “-omics” that allows investigation of translational regulation is known as translatome

2.1.4 Translatome: “What cells need to happen”

Translatome is the measure of genes translational efficiency on a global scale

A reliable measure of translational efficiency of cellular mRNAs is the degree

of association with ribosomes where actively translated mRNAs are typically bound by several ribosomes and referred to as polysomes (Eldad and Arava 2008) As discussed by Inaki (2011), polysome formation has been shown to reflect the translational state of a transcript as examined by Western blot and

35S-methione incorporation (Beilharz and Preiss 2004; Wang et al 2010) Therefore, translatome information can be obtained via a polysome extraction step that allows clustering mRNAs with respect to ribosome loading followed

by an mRNA quantification step with microarrays for global scale analysis or quantitative real time polymerase chain reaction (qRT-PCR) for specific mRNAs

Recently, a few techniques have been developed for polysome extraction based on affinity capture of ribosomes or nascent peptides A protein A-tagged version of the ribosomal protein RPL16 was expressed in yeast and endogenously formed ribosomes and polysomes were recovered from cellular extracts with anti protein A IgG coupled microbeads (Halbeisen et al 2009)

On the other hand, heat shock protein HSP70 which binds on nascent peptide

to assist in protein folding (Hansen et al 1994), was used as a molecular anchor for separating polysome-loaded mRNAs (with nascent peptide) from free mRNA (no peptide) by affinity beads capturing HSP70 (Kudo et al 2010) However, to date, the gold standard to monitor translational efficiency

is polysome profiling (Arava et al 2005; Arava et al 2003; Mašek et al 2011)

which allows separation of translated mRNA with respect to their ribosome

loading (density) through a sucrose gradient (Figure 2.3) The separation can

resolve the free subunits 40 and 60S, the 80S complex (non active ribosomes)

as well as the consecutive polysomes with increasing number of ribosomes (active ribosomes) Polysomes are then fractionated and ribosome bound mRNAs are extracted, purified and quantified by microarrays

Figure 2.3: General overview of the polysome profiling approach used to generate translatome data

Polysomes are first separated by density on a sucrose gradient Heaviest mRNAs (i.g higher number of ribosome loading) are found at the bottom of the centrifugation tube Then polysomes are profiled to display the distribution of ribosome loading where higher translational efficiency corresponds to higher level of polysomes For mRNA quantification, ribosome bound mRNAs are extracted from polysome profiles in several fractions of different translational efficiency, purified and quantified on microarrays to generate translatome data

The advantage of polysome profiling is that there is no need for pre-expression

of a protein A-tagged protein and the profiles account for the direct association

Translatome

mRNAs are probed on different microarrays based on their translation efficiency

Sucrose gradient Density

Polysome profiling

60S 80S

Monosomes

Polysome fractionation and mRNA extraction

between ribosomes and mRNAs as opposed to the binding of chaperone to nascent peptides Consequently, the most established technique of polysome profiling was selected for the thesis

Such translatome approaches with polysome profiling have been successfully employed and reported in several areas of research, but never in the context of CHO cell cultures For example, the translatome of normally growing bacteria confirmed the high diversity of translational states and showed that mRNAs of transcriptional regulators were highly translated (Picard et al 2012) In a cancer cell line, translatome data were generated to identify mRNAs that continue to associate with polysomes during hypoxia stress (Thomas and Johannes 2007) A similar translatome approach applied on exponentially growing yeast showed that majority of mRNAs were highly translated but not for 43 genes suggesting that they may be translationally controlled (Arava et

al 2003) In another study on yeast, translatome data were used to investigate the coordination between transcriptome composition and mRNA translation after heat shock or rapamycin treatments (Preiss et al 2003) The translatome

of stem cells was also investigated to understand the role of translational control during stem cells differentiation (Sampath et al 2008) Importantly, Hendrickson et al (2009) improved the approach to generate translatome data and showed that micoRNAs worked by inhibiting translation initiation or by stimulating ribosome drop-off in human embryonic kidney (HEK)-293 cells

Translatome of cells is regulated at the molecular level by translational control

2.2 Translational control

The following section will describe the significance of sophisticated translational control mechanisms, what they are, how they work and their relevance to CHO cell cultures

2.2.1 Reasons for regulating translational activity

As discussed by Mathews et al (2007) there are several reasons for developing translational control mechanisms, firstly due to energy conservation Translation of mRNA is a highly Adenosine Tri-Phosphate (ATP; energy) consuming process for cells, reported to sequester not less than 20% of total cellular ATP in rat thymocytes (Buttgereit and Brand 1995) Through translational control, cells ensure to translate only what they need to minimize wastage of precious energy Secondly, translational control allows cells to quickly implement desired change of protein levels as any control step prior to translation would require greater response time due to intermediate steps like mRNA synthesis, mRNA processing, mRNA transport from nucleus

to cytoplasm and inevitably entail delay Thirdly, translational control offers the possibility to regulate the site of protein synthesis in the cytoplasm in response to spatial requirements (Besse and Ephrussi 2008; St Johnston 2005) while transcriptional control is restricted to the nucleus for instance Fourthly, translational control provides cells with flexibility of regulation because there are various mechanisms (see section 2.2.4) to fine-tune the level of regulation desired Finally, translational control can easily be switched on and off as required as it is often mediated by reversible protein modifications such as phosphorylation